Importin-{beta} Family Members Mediate Alpharetrovirus Gag Nuclear Entry via Interactions with Matrix and Nucleocapsid

The retroviral Gag polyprotein orchestrates the assembly andrelease of virus particles from infected cells. We previouslyreported that nuclear transport of the Rous sarcoma virus (RSV)Gag protein is intrinsic to the virus assembly pathway. To identifycis- and trans-acting factors governing nucleocytoplasmic trafficking,we developed novel vectors to express regions of Gag in Saccharomycescerevisiae. The localization of Gag proteins was examined inthe wild type and in mutant strains deficient in members ofthe importin-ß family. We confirmed the Crm1p dependenceof the previously identified Gag p10 nuclear export signal.The known nuclear localization signal (NLS) in MA (matrix) wasalso functional in S. cerevisiae, and additionally we discovereda novel NLS within the NC (nucleocapsid) domain of Gag. MA utilizesKap120p and Mtr10p import receptors while nuclear entry of NCinvolves the classical importin- ${alpha}$ /ß (Kap60p/95p) pathway.NC also possesses nuclear targeting activity in avian cellsand contains the primary signal for the import of the Gag polyprotein.Thus, the nucleocytoplasmic dynamics of RSV Gag depend uponthe counterbalance of Crm1p-mediated export with two independentNLSs, each interacting with distinct nuclear import factors.

INTRODUCTION

Top

Introduction

Retroviruses, as obligate intracellular parasites, exist ina dynamic relationship with the host cell. These significantpathogens coopt host cell machinery to facilitate viral entry,gene expression, protein synthesis, and virion release. Identificationof cellular factors involved in the retroviral life cycle couldbe facilitated by the use of a well-characterized genetic system.Saccharomyces cerevisiae serves as the ideal candidate, as eachgene has been systematically deleted or mutated, enabling asearch for gene products that participate in viral processes.Although the complete retrovirus replication cycle cannot berecapitulated in yeast, discrete steps of the life cycle canbe studied in detail to identify host factors that are required.

The gag gene, one of the three major genes shared by all retroviruses,encodes the Gag polyprotein, which drives the formation andrelease of virus particles from the host cell membrane. Gagproteins are synthesized on cytosolic ribosomes, and previouslyit was thought that Gag was transported directly to the plasmamembrane to drive virus assembly. However, we discovered thatthe Gag protein of the alpharetrovirus Rous sarcoma virus (RSV),a tumor-producing virus that infects domestic fowl, undergoesnucleocytoplasmic shuttling prior to its transport to the plasmamembrane (36). The role of Gag nuclear trafficking in the virallife cycle is not well understood; however, both virion releaseand infectivity are inhibited when Gag nuclear export is prevented(36; L. Scheifele and L. Parent, unpublished data, 2004).

Following virion release, the RSV Gag polyprotein undergoesproteolytic cleavage into the MA, p2, p10, CA, NC, and PR proteins.Extracellular virus particles bind to receptors and undergofusion with the cell membrane, and the viral core enters thecytosol. Reverse transcription proceeds, and the viral ribonuclearcomplex (known as the preintegration complex [PIC]) enters thenucleus, where proviral DNA integrates into the cellular genome.For human immunodeficiency virus type 1 (HIV-1), which infectsnondividing cells, the PIC traverses through the nuclear porecomplex (NPC) by use of redundant nuclear localization signals(NLSs) in the integrase, Vpr, and MA proteins (40, 49). In contrast,RSV primarily infects dividing cells, and therefore it was believedthat the import of the PIC depended on the breakdown of thenuclear envelope during mitosis. However, RSV does replicateat low levels in quiescent cells (15, 18); thus, there is somelevel of active transport of the PIC, and nuclear entry likelydepends on the karyophilic properties of viral proteins.

Nuclear transport occurs through the NPC, which is composedof $~$ 30 nucleoporins (35), via interactions between NLSs on proteincargoes and a family of soluble nuclear receptors known as importinsor karyopherins (reviewed in reference 45). The classical importpathway involves direct recognition of a basic NLS by importin-ß(Kap95p) or the use of importin- ${alpha}$ as an adaptor for binding toimportin-ß. The diversity of nonclassical NLSs isstill being elaborated, and transport of proteins bearing nonclassicalimport signals is incompletely understood; however, karyopherinfamily members other than importin-ß appear to transportthese specialized cargoes.

Using S. cerevisiae strains bearing deletions or conditionalmutations in genes encoding essential or nonessential importin-ßfamily members, we determined that the route of nuclear entryfor the RSV MA protein, which contains a noncanonical NLS, dependson Mtr10p and Kap120p. We also identified a novel nuclear targetingactivity within the Gag NC domain that depends on the recyclingof importin- ${alpha}$ . Importantly, in avian cells (the natural hostfor RSV), the MA NLS had demonstrable but weak activity bothwhen expressed alone as a green fluorescent protein (GFP) fusionprotein and in the context of the Gag polyprotein. However,the NC protein exhibited stronger nuclear localization and appearedto be the major karyophilic signal for the nuclear import ofGag. Thus, the use of the powerful yeast genetic system ledto our discovery that independent NLSs in MA and NC utilizedistinct nuclear import receptors that may play critical rolesin the virus life cycle.

MATERIALS AND METHODS

Top

Materials and Methods

Yeast strains and methods. For initial studies, yeast strain W303 (MATa ade2-1 ura3-1 his3-1,115trp1-1 leu2-3,112 can1-100) was used. For analysis of nonessentialimportin-ß family members, strains from the yeastdeletion collection (Open Biosystems) were used (13). Strainsused to assess essential importin-ß family memberswere as follows: Y1706 (MAT ${alpha}$ ura3-52 ade2-101 his3-11 trp1- ${Delta}$ 901CSE1) and Y1709 (MATa ura3-52 ade2-101 his3-11 trp1- ${Delta}$ 901 cse1-1)(52), both from M. Fitzgerald-Hayes; PSY589 (MATa PSE1 ura3-52trp1 ${Delta}$ 63 leu2 ${Delta}$ ) and PSY1201 (MATa pse1-1 ura3-52 trp1 ${Delta}$ 63 leu2 ${Delta}$ 1);PSY 1213 (MATa pse1-7 ura3-52 trp1 ${Delta}$ 63 leu2 ${Delta}$ 1) (37); PSY 870 (MATaRSL1/KAP95 rna1-1 ade2 ade3 leu2 ura3 and carrying pPS714 witha wild-type [wt] RNA1) and PSY871 (MATa rsl1-1/kap95 rna1-1ade2 ade3 leu2 ura3 and carrying pPS714 with a wild-type RNA1)(19), all from P. Silver; NOY 388 (MATa SRP1/KAP60 ade2-1 ura3-1his3-11 trp1-1 leu2-3,112 can1-100) and NOY 612 (MATa srp1-31/kap60-310ade2-1 ura3-1 his3-11 trp1-1 leu2-3,112 can1-100) (53), bothfrom M. Nomura; DF5 ${alpha}$ (MAT ${alpha}$ KAP104 lys2-801 leu2-3,2-112 ura3-52his3 ${Delta}$ 200 trp1-1) and kap104-16 (MAT ${alpha}$ kap104::ura3::HIS3 lys2-801leu2-3,2-112 ura3-52 his3 ${Delta}$ 200 trp1-1 and containing pRS314kap104-16)from L. Pemberton (1); the W303 XPO1 (MATa xpo1::LEU2 ade2-1ura3-1 his3-1,115 trp1-1 leu2-3,112 can1-100 ura3-1 and containingplasmid pKW440 encoding XPO1) and W303 xpo1-1 (MATa xpo1::LEU2ade2-1 ura3-1 his3-1,115 trp1-1 leu2-3,112 can1-100 ura3-1 andcontaining plasmid pKW457 encoding xpo1-1) strains (43) fromC. Guthrie; and LLY1044 (MAT ${alpha}$ ${Delta}$ crm1::KAN^r leu2^– his3^–trp1^– ura3^– and containing a functional CRM1-HAon pRS315) and MNY8 (MAT ${alpha}$ ${Delta}$ crm1::KAN^r leu2^– his3^–trp1^– ura3^– and containing pDC-CRM1T539C encodinga leptomycin B [LMB]-sensitive Crm1p on pRS315), both from M.Rosbash. The MTR10 gene is unessential in the genetic backgroundused to construct the mtr10 ${Delta}$ ::Kan^r strain and was constructedby gene replacement as described previously (3).

Yeast growth and transformations followed standard procedures(39). For the induction of protein expression using pEMBLyex4and pIGin_A/out_A vectors, cells were transferred to media with2% galactose as the carbon source. Induction generally proceededfor $~$ 6 h, although induction of some constructs proceeded moreslowly. For analysis of protein localization in temperature-sensitivestrains, cells were grown at a permissive temperature (23°C)with 2% glucose, induced for 6 to 8 h in media with 2% galactose,and moved from permissive to nonpermissive temperatures (37°Cfor heat-sensitive mutations and 16°C for cold-sensitivemutations), and images were captured after 30 to 60 min at therestrictive temperature.

Plasmids. Plasmid constructions were performed in Escherichia coli strainDH5 ${alpha}$ using standard methods. The pIGin/out vectors (Fig. 1A)were created using pRS426 (New England Biolabs) (41) as a backbone.pIGin_A contained the GAL1 promoter to allow inducible gene expressiononly, while pIGin_B utilized the hybrid GAL10-CYC1 promoter topermit both constitutive and inducible expression. Histone 2B(H2B) sequences either with or without the nuclear localizationsignal were obtained using PCR amplification. Two tandem GFPsequences were amplified by PCR, with the insertion of a stopcodon after the second copy of the gene. Each PCR product wasligated into the pGEM-T (Promega) or Zero Blunt TOPO (Invitrogen)vectors, digested with the appropriate restriction enzyme, andligated into pRS426. Control sequences encoding NLSs from SV40large T antigen (22) and Mod5p-II (46) or nuclear export signals(NESs) from Gle1p and Rna1p (8) were inserted between BamHI-EcoRIsites in pIGin_A/B or pIGout_A/B. All gag-derived coding sequenceswere obtained from the wild-type RSV Prague C genome of pATV-8by use of PCR amplification, and fragments were subcloned intothe pIGin_A/B, pIGout_A/B, pEGFP.N2, or pECFP.C1 (Promega) vector.Plasmids pGag-GFP (4), pMA-GFP (11), and pGag. ${Delta}$ NC.GFP (5) havebeen previously described, and pYFP.NC was a kind gift fromE. Callahan and J. Wills. All plasmid sequences were confirmedby automated DNA sequencing.

View larger version (28K):
[in this window]
[in a new window]
FIG. 1. Use of novel yeast expression vectors for identification of nuclear targeting sequences. (A) Schematic diagram of S. cerevisiae inducible expression vectors. The pIGin_A and pIGout_A parental vectors, based on pRS426, are shown with relevant restriction sites indicated above. The GAL1 UAS promoter drives gene expression. pIGin_A contains codons 1 to 14 of the H2B sequence, which provides no subcellular targeting information, fused in frame to tandem copies of the GFP gene. pIGout_A is identical except for the presence of codons 1 to 67 of H2B, which includes the NLS. 2µ, 2µ origin of replication. (B) Live-cell imaging of yeast cells either uninduced (unind) or induced (ind) to express pIGin_A, pIGin_ANLS (SV40 T antigen NLS), pIGoutA, or pIGout_ANES (Rna1p NES). The yeast nucleus is indicated by an arrow.

Microscopic imaging. Yeast cells grown on solid medium were suspended in phosphate-bufferedsaline on glass slides and photographed using a Nikon Microphot-FXmicroscope with a 62x objective and a SenSys charge-coupled-devicecamera (Photometrics Ltd). Images were processed using QED software,with identical manual exposure settings used for experimentaland control cell imaging. Avian cells (transformed quail fibroblasts[QT6 cells]) were maintained in culture and transfected as describedpreviously (11), fixed with 2% paraformaldehyde, and imagedusing a Leica TCS SP2 AOBS confocal microscope at an excitationwavelength of 488 nm. Images were captured at appropriate microscopesettings so that no pixels were overexposed, and image processingwas performed using Corel software to make minor adjustmentsin the overall intensity of the gray scale images.

RESULTS

Top

Results

Creation of inducible vectors for subcellular localization studies. The objective of these studies was to identify the nuclear transportmachinery that interacts with the RSV Gag protein to facilitatenucleocytoplasmic trafficking. To this end, a Gag-GFP fusionprotein (4) was transferred into the pEMBLyex4 (47) and pIGin_Bvectors, which allow constitutive protein expression in additionto inducible expression via galactose regulation. However, expressionlevels of the fusion protein were very low, with distributionof the Gag-GFP protein either throughout the entire cell orwith a slightly enhanced nuclear pool, and some cells appearedenlarged or misshapen (data not shown). Although unfavorablecodon utilization could explain the low level of protein expression,attempts to induce higher expression levels had a paradoxicaleffect, with fewer cells expressing the Gag-GFP fusion protein,suggesting instead cellular toxicity. To limit the putativetoxic effects of Gag-GFP in yeast, we utilized vector pGP54a(a kind gift from J.E. Hopper), which carries only the GAL1promoter to allow for cell growth in the absence of the Gagprotein until its induction with galactose. Even under theseconditions, expression levels were very low, with <10% ofcells expressing Gag-GFP.

Because the difficulty in expressing the full-length Gag proteinlimited a comprehensive analysis of protein localization, wereasoned that smaller portions of Gag (MA, p10, and NC) mightbe used to identify domains of Gag that govern nucleus/cytosoldynamics. For these experiments, we developed a new inducibleexpression system for the assessment of protein location inlive cells (Fig. 1A). The pIGin_A/out_A vectors contain the 2µorigin of replication, allowing for multicopy plasmid replication( $~$ 5 to 20 copies per cell), with expression of the target genedriven by the inducible GAL1 promoter. The target protein wasinserted as a fusion with two tandem, in-frame copies of GFPto amplify the fluorescent signal and to prevent the fusionprotein from undergoing passive diffusion through the NPC. Theinserted protein sequence was not initiated from its own translationstart site but was cloned as a fusion with the yeast histone2B protein (23) because of the favorable translation initiationcontext. The pIGin_A vector contained the N-terminal 14 aminoacids of the histone 2B protein, which possesses no specifictargeting activity, to ascertain whether the inserted test sequencehas karyophilic properties. The pIGout_A vector contains thefirst 67 amino acids of histone 2B, which includes its NLS,to permit identification of test sequences that possess nuclearexport activity that counteracts the NLS and redirects the fusionprotein into the cytoplasm.

Control NLS or NES sequences were inserted into the appropriatepIGin_A/out_A parental vector to test the utility of the systemfor studying nuclear trafficking in live cells with fluorescencemicroscopy (Fig. 1B). Following induction, expression of thepIGin_A vector, which encodes tandem GFP proteins without a specifictargeting signal, revealed fluorescence confined to the cytoplasm.Insertion of the SV40 large T antigen NLS or the cellular Mod5pNLS (46) into the pIGin_A vector resulted in the localizationof the majority of the GFP fusion protein pool within the yeastcell nucleus (Fig. 1B, data shown for SV40 T antigen NLS). Thenucleus in budding yeast is located near the interface betweenthe mother cell and the daughter cell, and if cell divisionis not complete, nuclear contents can be seen streaming betweenthe cells (Fig. 1, pIGin_A NLS induced and pIGout_A induced).The location of the nucleus was confirmed by staining live cellswith DAPI (4',6'-diamidino-2-phenylindole) and by immunofluorescenceof fixed cells costained with DAPI and an antibody against GFP(data not shown).

Induced expression of the pIGout_A vector, which encodes tandemGFP proteins with the histone 2B NLS, revealed fluorescenceconfined to the nucleus (Fig. 1B). Addition of the Gle1p NES(8) to pIGout_A led to a partial redirection of the GFP signalfrom the nucleus into the cytoplasm, although a minor nuclearpool was apparent (data not shown). In contrast, the dual NESsof Rna1p (8) led to redistribution of the majority of the nuclearpool into the cytoplasm (Fig. 1B, pIGout_ANES). Western blotanalysis revealed that all of the proteins were stably expressedas predominantly full-length proteins, with negligible unconjugatedGFP detected (data not shown). Thus, the pIGin_A/out_A vectorsare effective reporters of NES and NLS activities in livingcells. Additional versions of the pIGin/out plasmids were generatedusing other selectable markers with and without centromericsequences and were found to function similarly.

Crm1p dependence of the p10 NES. To test whether these inducible expression vectors would beuseful for our studies of retroviral Gag targeting sequences,we examined whether the nuclear export activity identified previouslyin the p10 domain of RSV Gag (36) was a transferable signalrecognized by the transport machinery in S. cerevisiae. Initially,the entire p10 sequence was expressed using pIGout_A, which containsthe histone 2B NLS fused to two copies of GFP. Wild-type yeastcells induced to express pIGout_Ap10 displayed cytoplasmic localizationof the GFP reporter, suggesting that a sequence in p10 redirectedthe protein out of the nucleus (Fig. 2A, left panel). To identifythe amino acids in p10 that are sufficient for nuclear export,residues ₂₁₆GPALTDWARVREELAST₂₃₂ were substituted for the entirep10 sequence and expressed in the context of pIGout_A. This 17-amino-acidmotif comprising the p10 NES conferred cytoplasmic localizationto the nuclear reporter protein, suggesting that residues 216to 232 of p10 constitute a transferable NES that is active inwild-type yeast cells (pIGout_Ap10NES) (Fig. 2A, left panel).

View larger version (30K):
[in this window]
[in a new window]
FIG. 2. (A) Expression of pIGout_A, pIGout_Ap10, pIGout_Ap10NES, or pIGin_A in either the parental XPO1 strain or the conditional temperature-sensitive mutant xpo1-1 strain. Protein expression was induced for 6 h at 23°C with selective medium. Cells were incubated at 37°C for 30 min and imaged using fluorescence microscopy. (B) Summary of subcellular localization of GFP fusion proteins in the XPO1 and xpo1-1 strains. The experiment was performed as described for panel A, and then 100 individual cells were scored for subcellular localization of the fluorescent protein and placed into one of two categories: C>N (cytoplasm greater than nucleus) or N $≥$ C (nucleus greater than or equal to cytoplasm).

Although these results suggested that p10 residues 216 to 232exhibited nuclear export activity, it was possible that thelack of nuclear localization of pIGout_Ap10 and pIGout_Ap10NESwas due to a failure of the protein to undergo nuclear entry,despite the presence of the histone 2B NLS. Because the nuclearexport of Gag in avian cells is disrupted by LMB, a specificinhibitor of CRM1 activity (36), interfering with the Crm1ppathway in yeast would be predicted to cause an accumulationof the p10 or p10NES fusion proteins within the nucleus. InS. cerevisiae, Crm1p (also called Xpo1p or Exportin1) is essential,so a conditional, temperature-sensitive mutant strain was utilized(the xpo1-1 strain [43]). Fusion protein expression in transformedXPO1 (wild-type) and xpo1-1 cells was induced at 23°C, andcells were shifted to 37°C for 30 to 60 min and imaged withfluorescence microscopy (Fig. 2A). As expected, there was nodifference in the appearances of wild-type and xpo1-1 mutantcells expressing the pIGin_A or pIGout_A control proteins_. However,in the xpo1-1 strain, the localizations of pIGout_Ap10 and pIGout_Ap10NESwere dramatically altered from the cytoplasm to the nucleus(Fig. 2A, compare left and right panels).

To quantitate the pattern of nuclear/cytosolic expression inthe population of transformed cells, 100 individual cells werescored according to subcellular localization as predominantlycytoplasmic (C>N) or predominantly nuclear (N $≥$ C; this fractionincluded a small percentage of cells that expressed GFP equallyin the nucleus and cytoplasm). As shown in Fig. 2B, the controlprotein expressed by pIGout_A was nuclear in the wt (N $≥$ C, 98%)and xpo1-1 (N $≥$ C, 96%) strains. In contrast, pIGout_Ap10 and pIGout_Ap10NESfusion proteins were located in the cytoplasm in the majorityof wt cells (C>N, 78.6% and 81.7%, respectively), but eachprotein was nuclear in xpo1-1 cells (N $≥$ C, 80.5% and 70%, respectively).

Although the p10 NES operated through Crm1p when nuclear importwas directed by the NLS of histone 2B, we asked whether theNES would function properly in yeast when nuclear entry wasalso governed by the NLS in the Gag MA domain. To address thisquestion, the N-terminal portion of Gag containing the MA, p2,and p10 sequences was fused to tandem GFPs in the context ofthe pIGout_A vector. For pIGout_AMAp2p10 in wild-type XPO1 cells,36% of transformed cells demonstrated primarily nuclear localization,in contrast to the 98% of cells expressing pIGout_A, suggestingthat the p10 NES functioned in yeast in the context of the nativeGag sequence (Fig. 2B). In xpo1-1 cells expressing pIGout_AMAp2p10,there was an increase in the fraction of cells with nuclearlocalization (54.9%), confirming the Crm1p dependence of export.It is not clear why the percentage of xpo1-1 cells demonstratingfluorescence within the nucleus was less for pIGout_AMAp2p10than for pIGout_Ap10, although it is possible that the additionof the MA and p2 sequences influences the optimal exposure ofthe Gag NLS or NES in this context. Nonetheless, these dataconfirm that the p10 NES provides nuclear export capabilitieswhen in proximity to its normal cis-acting sequences in Gag.

To verify the results found with the xpo1-1 mutant, we alsoinhibited Crm1p with LMB using strain MNY8 (27), which bearsa mutant of Crm1p that is sensitive to LMB due to a C539T substitution,and a similar outcome was observed (data not shown). Taken together,these data indicate that the p10 NES functions in a Crm1p-dependentfashion that is recapitulated in S. cerevisiae, signifying thatthis is a relevant and informative genetic system for the studyof host factors that govern subcellular localization of RSVGag proteins.

Nuclear localization of MA and NC proteins using the pIGin_A/out_A vector system. To ascertain whether the nuclear targeting activity of MA thatwas previously identified in avian cells would be amenable tostudy with S. cerevisiae, the MA protein was fused in frameto the tandem GFP protein. Expression of pIGin_AMA revealed anaccumulation of the fusion protein in the nucleus with residualcytoplasmic localization (Fig. 3). This distribution of theMA-GFP protein is the same as that in avian cells (36), suggestinga similar targeting mechanism might be shared in lower and highereukaryotic cells. In a minor subpopulation of cells, a smallcytoplasmic dot distinct from the nucleus was observed in cellsinduced for longer periods of time ranging from 6 to 24 h. Thesecytosolic dots might represent misfolded MA-GFP fusion proteinsthat accumulate in the cytoplasm.

View larger version (67K):
[in this window]
[in a new window]
FIG. 3. Live cells expressing pIGin_A, pIGin_AMA, or pIGin_ANC. Yeast strain W303 was transformed and induced to express the indicated plasmid, and live cells were imaged $~$ 6 h following induction. Adjacent panels show two examples of images obtained for each transformant.

Despite evidence that MA possesses nuclear targeting activity,the sequence does not contain a classical motif of basic residuesthat is characteristic of either monopartite or bipartite NLSs.Unlike MA, the NC domain of Gag contains several clusters ofbasic residues, with 11 of 52 amino acids being Lys or Arg.Therefore, we postulated that NC might also possess karyophilicproperties. Consistent with this idea, expression of pIGin_ANCresulted in nearly complete nuclear localization of NC-GFP withvery little cytoplasmic signaling, indicating robust nucleartargeting activity (Fig. 3). Thus, we identified a novel NLSin NC, suggesting that the RSV Gag polyprotein might containtwo independent signals for nuclear entry.

Utilization of MA and NC NLSs for nuclear import of the Gag polyprotein in avian cells. Finding a new karyophilic signal in NC that is recognized bythe yeast nuclear transport machinery raised the possibilitythat this sequence might play a role in the admittance of Gagproteins into the nucleus during RSV infection. Although wepreviously observed (36) nuclear targeting activity of the RSVMA protein in avian cells, as shown in Fig. 4, the localizationof the NC protein in these cells had not been examined. To ascertainwhether the NLS in NC that we discovered in yeast was also activein avian cells, the sequence encoding NC was linked to yellowfluorescent protein (YFP) as a C-terminal fusion, and QT6 cellsexpressing YFP-NC were examined using confocal microscopy. Althoughthe distribution of YFP was throughout the entire cell, theYFP-NC protein was strongly localized to the nucleus and wasfurther concentrated within subnuclear structures identifiedas nucleoli (Fig. 4B and data not shown).

View larger version (45K):
[in this window]
[in a new window]
FIG. 4. Avian cells expressing RSV Gag derivatives. (A) Schematic diagram of proteins expressed as GFP or YFP fusions (shaded ovals). Wild-type RSV Gag-GFP is represented at the top with the following cleavage products indicated: MA (matrix), p2, p10, CA (capsid), NC (nucleocapsid), and GFP in place of PR (protease). Full-length MA fused to GFP and MA deletion mutants in Gag-GFP are illustrated, with the amino acids deleted within MA shown as subscripts. For ${Delta}$ NC Gag-GFP, seven amino acids at the N-terminal portion of NC remain, followed by GFP. Full-length NC is fused in frame at its C terminus with YFP, which is indicated by a darkly shaded oval. (B) QT6 cells were transfected with the indicated constructs, cells were fixed with 2% paraformaldehyde, and images were obtained by confocal microscopy. (C) QT6 cells were processed as described for panel B. Cells indicated as + LMB were incubated with 18 nM leptomycin B for 2 h prior to fixation.

Because MA and NC each possess nuclear import activity, we askedwhether both NLSs contribute equally to the nuclear entry ofthe Gag polyprotein or whether one might be the predominantsignal. To test this idea, constructs with deletions in theMA or NC domains of Gag-GFP were expressed in avian cells inthe presence or absence of LMB, and the degree of nuclear localizationwas assessed by confocal microscopy (Fig. 4C). Gag ${Delta}$ NC-GFP containsa C-terminal deletion of NC and removes all but two basic aminoacids as well as the zinc finger-binding domain. Like the full-lengthGag-GFP protein, Gag ${Delta}$ NC-GFP was located in the cytoplasm of untreatedQT6 cells, but upon exposure to LMB, a portion of the proteinpool became located within the nucleus (Fig. 4C) (5). However,there was more fluorescence present in the cytoplasm of LMB-treatedGag ${Delta}$ NC-GFP-expressing cells than in the cytoplasm of those expressingGag-GFP. The nuclear import of Gag ${Delta}$ NC-GFP presumably dependsupon the NLS in MA, and nuclear localization is incomplete.The data are consistent with NC contributing to the abilityof Gag-GFP to localize to the nucleus.

To determine whether deletion of the MA NLS would similarlyreduce the nuclear accumulation of Gag-GFP, two constructs withdeletions either of the N-terminal portion of MA ( ${Delta}$ MA_5-86 Gag-GFP)or of nearly the entire MA sequence ( ${Delta}$ MA_5-148 Gag-GFP) were examinedin QT6 cells (Fig. 4A). For both MA deletions, localizationwas cytoplasmic, with nuclear exclusion in untreated cells,although the larger deletion ( ${Delta}$ MA_5-148 Gag-GFP) resulted in amore punctate appearance (Fig. 4C). Treatment with LMB led tothe nuclear localization of the MA-deleted GFP fusion proteinto a degree similar to that for full-length Gag-GFP, indicatingthat the remaining NLS within NC was sufficient for the nuclearentry of Gag. Therefore, in the context of Gag, it appears thatthe NC domain contains a stronger NLS than MA, implying thatNC might be the dominant signal for nuclear translocation ofthe Gag polyprotein.

Identification of importin-ß family members utilized for import of MA and NC. We undertook these studies to identify both cis and trans elementsinvolved in the transport of RSV Gag proteins into the nucleus.Based on the absence of a canonical NLS motif within the MAsequence, we predicted that its nuclear import might dependon a trans-acting factor other than the classical importin-ß(Kap95p) transport pathway. To examine this question with anunbiased approach, we screened a deletion collection of nonessentialnuclear transport genes in S. cerevisiae to identify strainsthat were deficient in nuclear localization of pIGin_AMA (Fig.5 and Table 1). The MA fusion protein displayed enriched nuclearlocalization in all of the deletion strains except those strainslacking mtr10 or kap120, indicating that Mtr10p and Kap120pare important for MA nuclear entry. Because both Mtr10p andKap120p appear to facilitate the import of MA, we asked whetherthese importins might be involved in redundant pathways. Doublemutants generated by mating, sporulation, and dissection hadgrowth much poorer than that of either single mutant (data notshown). The "synthetically sick" phenotype of the mtr10 ${Delta}$ kap120 ${Delta}$ segregants suggests overlapping functions of these import factors.Of the essential importin-ß family members tested,xpo1-1 had no effect on localization. However, no conclusionscould be drawn about the roles of Cse1p, the transporter involvedin the cytoplasmic recycling of importin- ${alpha}$ (21, 42, 52) or Kap104pin pIGin_AMA localization, because at the nonpermissive temperaturethere were no nuclear pools in either the wt or the conditionalcse1-1 and kap104-16 mutant strains.

View larger version (38K):
[in this window]
[in a new window]
FIG. 5. Localization of GFP fusion proteins in strains carrying deletions or conditional mutations of genes encoding importin-ß family members. Plasmids used to transform cells are listed above, and characteristics of yeast strains are shown to the left. mtr10 ${Delta}$ and kap120 ${Delta}$ deletion strains were analyzed after the induction of protein expression at 23°C for $~$ 6 h. For the essential CSE1 gene, the parental CSE1 strain and the cse1-1 cold-sensitive strain were imaged following transfer of induced cells to 16°C for $~$ 30 min.

View this table:
[in this window]
[in a new window]
TABLE 1. Subcellular localization of fusion proteins in yeast strains containing deletions or conditional mutations of importin-ß family members

Analysis of pIGin_ANC localization in the importin-ßmutant strains revealed that none of the nonessential geneswas involved (Fig. 4 and Table 1). However, at restrictive temperatures,the cse1-1 mutant strain displayed reduced nuclear signalingand enhanced cytoplasmic fluorescence, implicating a role forthe classical importin- ${alpha}$ /ß pathway in the nuclear transportof NC. The fact that pIGin_ASV40 (SV40 large T antigen NLS),which utilizes importin- ${alpha}$ /ß for nuclear entry, alsohad increased cytoplasmic accumulation in the cse1-1 strainat the nonpermissive temperature verified that the mutant strainbehaved appropriately. The involvement of importin- ${alpha}$ (Srp1p),importin-ß (Rsl1p), or RanBP6 (Pse1p) could not bedirectly tested because of an inability to induce expressionof NC or MA fusion proteins in the corresponding conditionalmutant strains, most likely due to the dependence of the GALinduction system on the classical nuclear transport system (30).

DISCUSSION

Top

Discussion

The nuclear trafficking of RSV Gag was only recently discovered,and the possible role(s) Gag proteins play in the nucleus duringvirus assembly remains uncertain (36). However, nuclear entryof retroviral DNA during early infection is a well-describedfeature of all retroviruses and is vital to the establishmentof infection. The viral signals that mediate nuclear importat this critical early stage of infection are poorly characterizedfor RSV, yet it is likely that at least some of the mature Gagproteins are involved. In fact, we found that both the matureMA and NC proteins possess nuclear targeting signals. In aneffort to shed light on the nuclear transport pathways usedby these Gag proteins, we utilized genetic tools developed instudies with S. cerevisiae to identify importin-ßfamily members that mediate the import of MA and NC. Understandingthe cellular factors usurped by Gag proteins for nuclear entrymay shed light on the function(s) of Gag in the nucleus andmay begin to elucidate the molecular mechanisms involved inimport of the PIC.

The initial aim of studying the nuclear transport pathway ofthe Gag polyprotein was thwarted by its apparent cellular toxicityin yeast. We can imagine several possible mechanisms of Gagcytotoxicity. Gag proteins interface with the machinery associatedwith vacuolar sorting through multivesicular bodies during thelatter stages of virus assembly (reviewed in reference 24).Because the rigid yeast cell wall precludes viral budding, itis possible that Gag proteins accumulate in vacuolar bodiesand impair normal functions of the vacuolar sorting machinery.As well, Gag proteins targeted to the yeast cell membrane mightinterfere with cell growth and division. Both of these ideasare consistent with the elongated appearance of yeast cellsexpressing Gag proteins. Alterations in cellular morphologyand toxicity are also observed in avian cells that express highlevels of Gag-GFP, and the degree of cytotoxicity is more evidentwith Gag mutants that are strongly plasma membrane associatedthan in those mutants having a cytosolic distribution (5, 11,36). Furthermore, the nuclear trafficking of Gag might itselfbe toxic to yeast, particularly if the import factors used byGag are needed for other cellular functions. Efforts to isolatespontaneous mutants of yeast that are able to tolerate highlevels of Gag expression may prove to be informative for identifyingadditional host factors that interact with the virus assemblypathway.

S. cerevisiae serves as a model system for nucleus/cytosol exchangebecause the nuclear transport pathways are very highly conservedamong lower and higher eukaryotes (7). We constructed a setof inducible expression vectors to identify cis- and trans-actingfactors involved in Gag nucleus/cytoplasm dynamics in live yeastcells. Using these new pIGin/pIGout vectors, our data demonstratedthat residues 216 to 232 in the p10 domain of Gag are sufficientto mediate nuclear export through the Crm1p pathway. The NLSin the MA protein we previously described (36) was also confirmed,and we discovered a novel NLS within the NC domain. Verificationthat the NC NLS was functional in avian cells validated theuse of the yeast expression system to identify new targetingelements in heterologous viral proteins.

There are examples of proteins that possess multiple NLSs, someof which are served by different importin-ß familymembers (25), but in general, little is known about how eachindividual signal is utilized. Accordingly, we wonder why thereare two independent NLSs in the RSV Gag polyprotein. One possibilityis that both signals are used during the nuclear traffickingof Gag during virus assembly, and the presence of redundanttargeting signals in Gag would imply that nuclear traffickingis an important step in replication. However, our experimentssuggest that the novel NLS in NC is the predominant sequenceused for nuclear import of Gag in avian cells. Another possibleexplanation for the dual import signals is that the MA and NCNLSs might function either at separate steps or within distinctimport complexes to facilitate Gag nuclear entry. Finally, perhapseither the MA or NC NLS has no function in nuclear entry, sincethe basic residues in NC are also important for viral RNA bindingand the basic residues in MA mediate plasma membrane targeting.

It is also conceivable that the NLSs in MA and/or NC are activeduring the early steps of infection, when mature Gag proteinsenter the cell with the incoming virion. It is uncertain whetherMA is a component of the RSV PIC, although genetic evidencesupports this idea (12). Most certainly, NC is present withinthe PIC because it is bound to the RNA genome in the virionand is involved in reverse transcription and integration (2,6, 16, 34, 48). An NLS in RSV integrase was previously identified(20), and thus there are three candidate karyophilic proteinsthat might promote nuclear transport of the viral genome earlyin infection. The nuclear import of the HIV-1 PIC has been intensivelystudied, and it appears that there are multiple signals fornuclear targeting, including MA, integrase, Vpr, and an intermediateDNA product of reverse transcription (40). As well, HIV-1 NCenters the nucleus early after infection and enhances the expressionof spliced viral mRNAs after integration (54).

Interestingly, our screen of deletion strains and conditionalmutants involving importin-ß family members indicatedthat the MA and NC proteins enter the nucleus through interactionswith different soluble nuclear transport receptors. Import ofthe MA protein was dependent on Mtr10p and Kap120p. Mtr10p (mRNAtransport regulator) is involved in nuclear translocation ofRNA-binding proteins that contain serine/arginine (SR)-richdomains, including Np13p, Gbp2p, and Hrb1p, which are nucleocytoplasmicshuttling proteins implicated in the nuclear export of mRNAin yeast (14, 26, 29, 38, 50). These shuttling SR proteins areimported by Mtr10p and recruited to mRNA during transcriptionand then accompany the ribonuclear complex into the cytoplasm,where they may regulate translation (51). The dual involvementof Mtr10p in the import of MA and in the import of shuttlingmRNA binding proteins is intriguing because we have postulatedthat RSV Gag enters the nucleus to bind newly transcribed viralRNA and transport it into the cytoplasm for encapsidation intoassembling virus particles (36). However, there is no identifiablesequence homology between MA and SR proteins, including theNLS in Npl3p (10, 38; also, data not shown). Because littleis known about the substrate specificity or molecular interactionbetween cargoes for the human (transportin-SR or importin 12[17]) or avian homologues of Mtr10, RSV MA may prove to be animportant future tool for studying this import pathway in highereukaryotes.

We also found that the deletion of Kap120p, proposed to be abidirectional transporter through the NPC (9, 44), impairedMA nuclear import. Kap120p functions in the export of the 60Sribosomal subunit (44) and its human homologue, importin 11,has been implicated in the import of ribosomal protein L12 (31,32). Importin 11 also serves as the import receptor for theubiquitin-conjugating enzyme UbcM2, and covalent attachmentof ubiquitin to UbcM2 is a prerequisite for nuclear transport(33). This is an intriguing observation, given the dependenceof RSV particle assembly on ubiquitination (28), prompting usto speculate that ubiquitin modification of Gag might be involvedin its nuclear translocation.

Because deletions of both mtr10 and kap120 interfered with thenuclear accumulation of MA, we proposed that these importinsmight be involved in redundant pathways. Indeed, double mutants(mtr10 ${Delta}$ kap120 ${Delta}$ ) grew very poorly, suggesting overlapping functionsfor these import receptors. Future studies of MA-Kap120p interactionsmay reveal new insights into the substrate specificity and structuraldeterminants of this poorly understood import receptor pathway.

In contrast to MA, the NC protein has a basic motif that resemblesthe canonical NLSs, and accordingly, its nuclear entry appearedto be mediated by the classical importin- ${alpha}$ /ß pathway.It will be of interest to identify the specific residues inNC that are sufficient to mediate nuclear entry. Furthermore,we do not yet know whether there is a direct or indirect interactionbetween NC and the avian homologue of importin- ${alpha}$ . Finally, therelative reliance of Gag nuclear entry on the avian Kap120p,Mtr10p, or importin- ${alpha}$ /ß homologues will be worthwhileto pursue.

The results of these studies raise many intriguing questions.Why does the Gag polyprotein contain two independent NLSs thatutilize distinct import receptors? Are the NLSs in MA and NCthat mediate Gag nuclear entry also involved in the nucleartargeting of the PIC and the entry of the incoming viral genomeduring early infection, or do they have other intranuclear roles?Are there cellular proteins that compete with RSV MA for utilizationof Mtr10p and Kap120p in avian cells? The use of the S. cerevisiaegenetic system has provided the initial key insights neededto address these compelling questions, which are critical forunderstanding the molecular interactions of RSV with its intracellularenvironment.

ACKNOWLEDGMENTS

We thank M. L. Whitney for construction of mtr10::Kan^r; J. E.Hopper, M. Fitzgerald-Hayes, P. Silver, M. Nomura, L. Pemberton,C. Guthrie, M. Robash, E. Callahan, and J. Wills for reagents;Scott Kenney for critical review of the manuscript; and thePenn State College of Medicine Core Facility for assistancewith DNA sequencing and confocal microscopy.

We greatly appreciate support from the NIH (grants R01CA76534to L.J.P. and R01GM27930 to A.K.H.), the NSF (predoctoral fellowshipto L.Z.S.), the Penn State College of Medicine Dean's FeasibilityAward (L.J.P. and A.K.H.), and the M.D. Research FacilitationAward (L.J.P.). This project was funded, in part, under a grantwith the Pennsylvania Department of Health using Tobacco SettlementFunds, and the Department specifically disclaims responsibilityfor any analyses, interpretations, or conclusions.

FOOTNOTES

* Corresponding author. Mailing address: Division of Infectious Diseases HO36, Penn State College of Medicine, 500 University Drive, Hershey, PA 17033. Phone: (717) 531-3997. Fax: (717) 531-4633. E-mail: lparent{at}psu.edu.

These authors contributed equally to this work.

Present address: Department of Molecular Biology and Genetics,Johns Hopkins University School of Medicine, 733 North Broadway,Baltimore, MD 21205.

REFERENCES

Top